RLWRLD, a South Korean startup founded in 2024, is developing a foundational AI model specifically designed for robotics. This model integrates large language models with traditional robotics software to enable robots to perform agile movements and exhibit logical reasoning. The company’s goal is to automate manual processes by allowing robots to learn and replicate human expertise, thereby enhancing efficiency in various work environments.
The startup recently emerged from stealth mode, securing 21 billion KRW (approximately $14.8 million) in seed funding. This round was led by venture capital firm Hashed, with participation from Mirae Asset Venture Investment and Global Brain. Strategic investors include prominent companies from Japan, Korea, and India, such as Ana Group, PKSHA, Mitsui Chemical, Shimadzu, KDDI, LG Electronics, SK Telecom, and Amber Manufacturing.
RLWRLD plans to utilize the seed funding to conduct proof-of-concept projects with its strategic investors, acquire computing infrastructure like GPUs, purchase robots and data collection devices, and recruit top research talent. A key focus is developing advanced five-finger hand movements, a capability not yet demonstrated by competitors like Tesla, Figure AI, and 1X.
The company is also collaborating with its strategic partners to explore automation of various human-centric workflows using its AI model. They are preparing a humanoid-based autonomous action demonstration scheduled for later this year. Additionally, RLWRLD aims to develop a platform supporting various robot types, including industrial, collaborative, autonomous mobile robots, and humanoids.
Founded by Jung-Hee Ryu, RLWRLD is his third startup. His previous ventures include Olaworks, acquired by Intel in 2012, and Future Play, a startup accelerator focusing on deep tech companies. Ryu was motivated to start RLWRLD after observing the rapid growth of AI startups in the U.S., Europe, and China, contrasted with a relative absence in Korea and Japan. He engaged with over 30 AI professors from Korea and Japan to understand their challenges and opportunities, leading to the formation of RLWRLD with six professors from top South Korean institutions like KAIST, SNU, and POSTECH.
While other companies like Skild AI, Physical Intelligence, Tesla, Google DeepMind, and Nvidia are also developing foundational models for robotics, RLWRLD believes it has an advantage due to its team of AI and robotics experts and access to high-degree-of-freedom (DoF) reference robots. This access is expected to result in superior performance outcomes compared to competitors relying on low-DoF robots, such as two-fingered grippers.
With strategic investors providing proximity to manufacturing sites, RLWRLD can rapidly gather valuable data to train its models. In 2024, Japan and South Korea collectively accounted for 9.2% of worldwide manufacturing production, offering a rich data environment for RLWRLD’s initiatives. The company aims to generate revenue through proof-of-concept projects and collaboration demonstrations with strategic partners, focusing initially on industrial applications due to their high demand for automation and willingness to invest.
RLWRLD currently has a team of 13 employees.
